Which statement explains how to control the familywise error rate when making multiple comparisons?

• A. Increase the alpha level for all tests.
• B. Ignore multiple tests and report unadjusted p-values.
• C. Use a post-hoc power analysis.
• D. Apply a method such as Bonferroni or Holm to control the familywise error rate.

Answer: (D)

Controlling the familywise error rate when making multiple comparisons requires adjusting the criteria for significance across all tests. The idea is to keep the probability of making at least one false positive somewhere in the set of analyses at a tolerable level.

Using a correction method like Bonferroni or Holm achieves this by tightening the criteria for each individual test. Bonferroni divides the overall alpha by the number of tests, so each test has a smaller chance of yielding a false discovery. Holm takes a stepwise approach: it orders the p-values and compares them to progressively less stringent thresholds, which often gives more power than Bonferroni while still protecting the overall error rate. Together, these methods ensure that the chance of at least one erroneous claim across all comparisons remains controlled.

Why the other ideas don’t fit: increasing the alpha level for every test raises the likelihood of false positives across the board. Ignoring the issue and reporting unadjusted p-values gives no protection against accumulating errors when many tests are run. A post-hoc power analysis is about estimating the study’s ability to detect effects after the fact, not about controlling the error rate across multiple tests.